Objectives. We sought information about the cost-related underuse of medications—which medications are underused, by whom, and how often.

Methods. Chronically ill adults were asked to identify how often they underused prescription medication for 16 health conditions because of the cost.

Results. Eighteen percent of respondents cut back on medication use owing to cost in the previous year, and 14% used less medication at least monthly. Although rates of underuse varied substantially across treatments, prescription coverage and out-of-pocket costs were determinants of underuse across medication types.

Conclusions. Many chronically ill adults frequently cut back on medications owing to cost. Patients are selective about the treatments they forgo. Out-of-pocket costs and inadequate prescription coverage may lead to adherence problems for many important medication types.

Many chronically ill patients take less of their medication than has been prescribed, owing to cost concerns, especially those patients with low incomes, multiple chronic health problems, or no prescription drug coverage.1–3 The consequences of cost-related medication underuse include increased emergency department visits, psychiatric admissions, and nursing home admissions, as well as decreased health status.4–6

Although the public health significance of cost-related medication underuse is becoming clear, we have only a limited understanding of how medication costs affect individuals with differing clinical and socioeconomic characteristics. Recent survey-based studies1,2 asked respondents to report on cost-related adherence problems without specifying individual treatment types. As a result, investigators have had only a limited ability to identify which medications patients are most likely to underuse. Approximately 73% of older adults who use prescription medications use more than 1, and 29% use 4 or more.7 Although the clinical significance of cost-related underuse is different when patients cut back on treatment for life-threatening asymptomatic illnesses (e.g., hypertension) versus disabling symptomatic conditions (e.g., back pain), it is unclear whether those who restrict their medication use do so uniformly or are selective in which treatments they forgo. In addition, previous studies that used global measures assume that individuals reporting cost-related underuse are describing a regular practice, rather than isolated events. However, there has been little research to determine whether this assumption is accurate.

The paucity of data on cost-related medication adherence problems has important implications not only for estimating their clinical significance, but also for understanding the extent to which adherence problems vary across socioeconomic groups. For example, non-White patients may be more likely than Whites to discontinue pharmacotherapy for depression,8 and thus may be especially likely to forgo antidepressant treatment when facing cost pressures. However, general measures of medication underuse may miss problems specifically associated with antidepressants or other therapeutic categories. In addition, global estimates of the relationship between out-of-pocket costs and underuse may mask important differences across treatments. Individuals may be more sensitive to cost pressures when taking medications that they perceive as having little impact on their health or longevity, while patients who believe that their treatment is essential might maintain high levels of adherence despite high costs.

For all of these reasons, policymakers and clinicians need more extensive information about chronically ill patients’ cost-related underuse of prescription medications. Accordingly, we surveyed chronically ill adults to ascertain the prevalence, frequency, and risk factors for cost-related medication adherence problems. The study builds on previous research by describing (1) the relationships among global measures of cost-related medication underuse, measures documenting underuse of specific treatments, and measures identifying individuals who frequently experience adherence problems; and (2) variation in the importance of socioeconomic risk factors for underuse across treatment types.

Study Participants

The study was based on surveys completed in November and December 2002 by a nationwide panel of adults living in the United States. The panel was instituted in 1999, with continual enrollment since that time to replace dropouts and to maintain appropriate representation across sociodemographic groups. Panel members were identified and recruited with random-digit dialing and a sampling frame consisting of all US households with an assigned telephone number. Potential panel members were offered free WebTV and Internet access in exchange for completing short on-line surveys several times per month. At the time of this study, the panel recruitment rate was 48% and it included more than 40 000 members. Researchers have compared the panel’s characteristics to the US Census Bureau’s Current Population Survey, the National Health Interview Survey, and an independent random-digit dialing sample. On most sociodemographic parameters (e.g., race/ethnicity and gender), key health behaviors (e.g., smoking), and the prevalence of chronic illnesses, the panel has consistently been found to be within a few percentage points of other national estimates.9,10 However, compared with census data, the panel has more married individuals (64% vs 60%), fewer households with incomes > $75 000 (18% vs 25%), and among those ≥ 50 years of age, fewer individuals with at least some college education (55% vs 59%).11,12

We used information from panel members’ enrollment surveys to identify all 5644 individuals aged 50 and older who reported that they used prescription medication for diabetes, depression, heart problems, hypertension, or high cholesterol. After 3 personalized e-mail requests, 4264 people (76%) completed the on-line informed consent form and questionnaire. Of these, 185 participants were excluded from the current analyses because they reported in their survey that they were no longer taking medication for any of the 5 index conditions, and 24 additional respondents were excluded because they were missing data on income. The final sample included 4055 respondents.

Among individuals sampled for the current study, respondents were more likely than nonrespondents to be White (88% vs 81%, P < .001), older (mean age 65 vs 63 years, P < .001), and to have some college education (66% vs 60%, P < .001 ). Respondents and nonrespondents were similar with regard to gender (P = .29), home ownership (P = .44), marital status (P = .16), and income (P = .41). In all analyses, we used poststratification weights to correct the distribution of respondents to match the distribution of the US population on age, sex, race/ethnicity, education, region, and metropolitan residence.13

Survey Items and Variable Creation

Respondents reported whether they used prescription medication for each of 16 chronic conditions: arthritis; asthma; chronic back pain or sciatica; high cholesterol; chronic obstructive pulmonary disease (chronic bronchitis, emphysema, or COPD); depression; diabetes; heartburn, acid reflux, or irritable bowel syndrome; atherosclerosis (blocked arteries in the heart, angina, or chest pain from heart disease); heart failure; high blood pressure or hypertension; myocardial infarction (heart attack); migraine headache; osteoporosis; stomach or duodenal ulcers; and stroke. On a condition-by-condition basis, participants were asked: “In the past 12 months, have you ever taken less of this medication than prescribed by your doctor because of the cost [emphasis in the survey]?” Respondents reporting cost-related underuse were asked: “Thinking specifically about your [condition] medication, in the past 12 months, how often did you do each of the following, because of the cost?” Using a 5-point ordinal response set in which 1 = “at least once a week” and 5 = “never,” respondents reported the frequency with which they “took fewer pills or a smaller dose,” “did not fill a prescription at all,” “put off or postponed getting a prescription filled,” “used herbal medicines or vitamins when [they] felt sick rather than take [their] prescription medication,” and “took the medication less frequently than recommended to ‘stretch out’ the time before getting a refill.”

We created 4 measures of cost-related medication underuse. The first measure was treatment-specific and identified all respondents who reported cost-related underuse associated with each medication type. The second measure identified individuals reporting some type of underuse (e.g., “taking fewer pills or a smaller dose”) for each medication type at least once per month. The third measure identified respondents who reported any cost-related underuse in the previous year (i.e., for any of their medications). This measure is similar to those used in previous survey-based studies.1,2 Finally, the fourth measure identified individuals reporting some form of underuse for 1 or more of their medications at least once per month.

The socioeconomic variables we examined as possible predictors of cost-related adherence problems were participants’ race/ ethnicity (White vs non-White), age, gender, education level (high school degree or less vs some college or more), and annual household income (< $20 000, $20 000–$39 999, $40 000–$59 999, or ≥ $60 000). We also examined 3 indices of respondents’ out-of-pocket medication cost pressures: total number of current prescription medications (1–2, 3–6, or ≥ 7), monthly out-of-pocket prescription drug costs ($0–$50, $51–$99, or ≥ $100), and whether the respondent reported having prescription drug coverage (any vs none).

We calculated the prevalence of cost-related medication underuse using each of the 4 measures for the sample as a whole and among respondents who used the 16 specific medication types. We then fit logistic regression models predicting cost-related underuse for the 10 most common treatments as a function of respondents’ race/ethnicity, gender, age, education, income, number of prescriptions, total monthly out-of-pocket medication costs, and prescription drug coverage. Because the primary mechanism by which prescription coverage might affect adherence problems is by lowering out-of-pocket costs, we constructed an alternate set of models controlling for all of the same covariates with the exception of out-of-pocket costs. These alternate models more fully capture the association between reported benefits and underuse. The study was underpowered to construct multivariate models predicting cost-related adherence problems associated with less common medication types and to examine multivariate predictors of underuse occurring at least once per month. All analyses were conducted with Stata version 8.1.14

Respondent Characteristics

Twenty-four percent of respondents (weighted) reported 7 or more prescription medications (Table 1). Although 83% of respondents (weighted) reported some prescription drug coverage, half of the weighted sample had monthly out-of-pocket medication costs of at least $51, and 25% had monthly costs of $100 or more. As shown in Table 2, the most common medications were for treating hypertension (70%), high cholesterol (50%), arthritis (28%), heartburn (25%), diabetes (24%), and depression (21%).

Participants reporting monthly medication costs of $100 or more were more likely than those with lower costs to report no prescription drug coverage (33% vs 11%, P < .0001). Respondents who had incomes < $20 000 per year were more than twice as likely to report no medication coverage than those with incomes of at least $60 000 (21% vs 8%, P < .0001). Compared with younger participants, those 65 years or older were more likely to report no prescription drug coverage (23% vs 10%, P < .0001), monthly medication costs of $100 or more (28% vs 23%, P = .007 ), and annual incomes < $20 000 (25% vs 20%, P = .03). Ninety-three percent of respondents reported some form of health insurance coverage, including private insurance (67%), Medicare (59%), Medicaid (12%), or use of Veterans Affairs health care facilities (9%).

Prevalence of Cost-Related Underuse Based on Different Measures

Overall, 18% of respondents reported at least one episode of cost-related medication underuse in the previous year (Table 2). Global rates of underuse ranged from 16% among respondents with prescriptions for antihypertensive medication to 36% among respondents with prescriptions for medication treating stomach ulcers. The majority of respondents taking each condition-specific medication reported cost-related underuse of that specific treatment type, although many reported cutting back only on medications for other conditions. For example, 20% of respondents who used asthma medication reported under-using that treatment, while 32% of these respondents reported under-using some medication in the previous year. Cost-related adherence problems were most common for medications treating arthritis, depression, back pain, asthma, migraine headaches, and stomach ulcers (all treatment-specific restriction rates ≥ 20%).

When examining the frequency of medication underuse, we found that 14% of all respondents (78% of those who reported cutting back on medication owing to cost) were under-using some medication at least once per month. At least 10% of respondents who used prescription medication for arthritis, heartburn, depression, back pain, COPD, asthma, migraine headaches, and stomach ulcers cut back on their use of those specific medications at least monthly. More than 25% of respondents taking medication to treat depression, back pain, COPD, asthma, migraine headaches, stomach ulcers, and stroke restricted use of 1 or more of their medications at least monthly.

Sociodemographic Predictors of Cost-Related Underuse

The general pattern of risk factors for cost-related underuse was similar across the 10 most common medication types (Table 3). Controlling for out-of-pocket medication costs and other socioeconomic variables, respondents older than 65 years of age were consistently less likely to cut back on medication use owing to cost than those between ages 50 and 54. Respondents’ race/ethnicity, gender, and educational attainment were not significantly associated with underuse in any of the multivariate models. Odds ratios for these covariates, as well as all 95% confidence intervals, are available from the corresponding author on request.

Respondents’ incomes and monthly out-of-pocket medication costs were each independently associated with cost-related underuse of treatments for numerous health problems. For 9 of the 10 most common medication types, respondents with annual incomes of less than $20 000 had higher relative odds of cost-related adherence problems than those with incomes of at least $60 000. Statistically significant odds ratios comparing lower- to higher-income respondents ranged from 3.0 (for antidepressants) to 6.4 (for medication treating heart disease). Respondents with monthly out-of-pocket medication costs of $100 or more had greater odds of underuse for 9 of the top 10 medication types, than respondents with monthly costs of $0–$50. Statistically significant odds ratios associated with high out-of-pocket costs ranged from 2.9 (for arthritis medications) to 4.6 (for medication treating hypertension).

Controlling for out-of-pocket costs, respondents who reported no prescription drug coverage were significantly more likely to report cost-related underuse of medications for hypertension, hyperlipidemia, heartburn, and back pain (all P < .05). As expected, the association between prescription drug coverage and cost-related adherence problems was greater when estimated based on the alternate multivariate models that did not adjust for out-of-pocket costs. Specifically, respondents without prescription coverage had at least 3 times greater odds of reporting underuse of medications treating hypertension, heartburn, coronary artery disease after a heart attack, and back pain; and patients without prescription drug coverage had 1.9 to 2.8 times higher odds of reporting underuse of all remaining medications shown in Table 3. P-values for adjusted odds ratios in the alternate models were < .05, except in the model predicting underuse of medication treating osteoporosis (P = .16).

We found that medication underuse owing to cost is common among older adults with such chronic illnesses as asthma, heart failure, and depression, for which inadequate adherence can quickly lead to serious health problems. Cost-related medication underuse was also reported often among individuals taking medications to prevent adverse events during a longer term, such as treatments for high blood pressure, high cholesterol, and diabetes. Moreover, this study confirms that most patients (78% across medication types) who report cost-related medication adherence problems are consistently under-using prescribed medication, rather than describing relatively infrequent events.

From a public health perspective, these data suggest that more than 1 million of the roughly 11 million adults diagnosed with diabetes15 (i.e., 11%) may be taking less hypoglycemic medication than prescribed because of the cost, and 750000 people with diabetes may be cutting back on their hypoglycemic medication at least once per month (7%). Similarly, 2.9 million of the 14.1 million American adults with asthma16 (20%) may be cutting back on their asthma medication because of cost pressures, and 1.6 million may be cutting back on their asthma medication on a monthly basis (12%). Given the importance of medication adherence for these and other chronic illnesses, interventions that reduce rates of cost-related medication underuse may achieve substantial reductions in chronic disease morbidity at the community level.

From a clinical perspective, these analyses suggest that cost-related adherence problems should be considered as a possible explanation whenever patients fail to respond to pharmacotherapy. When the health of individuals who underuse their medication owing to cost does not improve, typical clinician responses, such as increasing their dose or adding augmentation therapy, are unlikely to improve outcomes. Unfortunately, clinicians may interpret “medication failures” as an indication of patients’ lack of commitment to the treatment plan or lack of efficacy for the specific agent. As a result, clinicians may inadvertently exacerbate the underlying problem by discarding the patient’s prescription before all of the medication is used to try another drug—a waste that many patients can ill-afford. In contrast, identifying adherence problems resulting from cost pressures may lead to more effective remedies, such as switching to generic or lower-cost branded drugs, providing more doses to lengthen the time between refills, or providing information on medication payment assistance programs. Even if patients’ treatment costs cannot be reduced, clinicians should explicitly inquire about potential adherence problems owing to out-of-pocket costs and discuss with patients the importance of adhering to their prescribed regimen. Unfortunately, studies suggest that these discussions frequently do not occur.17,18

Many respondents with a given health problem reported cutting back on some of their prescription medications but not the ones to treat that condition. This suggests that individuals are selective about their treatment use (and underuse). Future research should seek to elucidate the treatment-specific factors that patients facing cost pressures weigh when deciding to forgo medication use. Even in the absence of comprehensive payment reforms, there may be important opportunities for clinicians to collaborate with patients in making the most sensible choices about their medication use, given their economic circumstances, values, and health status.

In multivariate models, we found that both respondents who used life-sustaining therapies and those who used medications to relieve symptoms were susceptible to adherence problems owing to out-of-pocket medication costs. However the magnitude of the risk gradient varied across medication types. Some treatments may have relatively little association with out-of-pocket costs because individuals are willing to forgo therapy even when costs are low; for other treatment types, individuals may resist even high cost pressures because they believe their medication is worth the investment. Regardless of the reasons, these data suggest that similar levels of prescription benefits will affect patients differently depending on their diagnosis.

Although this study provides a more detailed picture of cost-related medication underuse than many previous studies, it still has important limitations. Studies suggest that individuals are less influenced by social desirability bias when reporting information to a computer than during an in-person or written survey;19–21 it is still possible, however, that social desirability or recall bias may have led some respondents to underreport or overreport their adherence problems. Individuals taking multiple medications may have difficulty reporting their adherence behavior for specific treatment types or may not accurately associate treatments with their various medical conditions. Finally, potential respondents who were non-White and had less education were somewhat less likely to complete this survey. Although we adjusted for these differences with sampling weights, nonresponse by patients with lower socioeconomic status may have led to an underestimation of the prevalence of cost-related adherence problems, particularly for diseases that are associated with social class and race/ethnicity (e.g., asthma, diabetes, and hypertension).

Measurement problems associated with respondents’ reports of their prescription drug coverage may be of particular concern in the current study. Individuals often have a limited understanding of their health insurance benefits and may underestimate their coverage for prescription drugs.22,23 Moreover, respondents’ benefit packages (e.g., annual caps or the size of copayments) undoubtedly varied substantially in this nationwide sample. Finally, individuals with prescription drug coverage in the current market may be sicker than other study participants and may have a greater need for medications. For all of these reasons, the odds ratios shown in Table 3 are likely to underestimate the potential impact of expanding prescription drug payment programs on rates of cost-related medication adherence problems.

Participants in the current study were identified through a nationwide panel with characteristics that are similar to those of other large nationally representative studies. Nevertheless, the process of panel member identification, recruitment, retention, and completion of the current survey could have introduced biases into these results. If panel participants were more comfortable using the Internet than non-participants, and if this tendency was associated with a greater proclivity to follow their medication regimen as recommended by their clinicians, the current study’s estimates of the prevalence of cost-related underuse may underestimate the magnitude of this problem. Also, participants self-identified as having one of the 5 chronic illnesses determining eligibility (as well as the other illnesses in the survey), and these reports were not validated through medical record reviews. We expect that individuals who did not identify that they had a given diagnosis would have had less severe conditions, may have been less likely to be taking prescription medications for that illness, and therefore would be less subject to adherence problems owing to cost.

In summary, we found that large numbers of chronically ill adults in this national study had taken less medication than prescribed during the previous year because of the cost. Most seriously ill individuals, such as those with heart failure or asthma, who reported cost-related adherence problems were cutting back on medication to treat those illnesses. However, individuals did not cut back uniformly on their treatments, and large numbers of respondents reported limiting their medication underuse to medications for other conditions. Most respondents reporting cost-related adherence problems were cutting back frequently, which could result in significantly less medication use than prescribed. Although the relationship between out-of-pocket costs and an individual’s propensity to use less treatment varied across medication types, cost and the availability of prescription drug coverage affected a patient’s adherence both to medications treating life-threatening conditions and to medications primarily treating symptoms. Increasing prescription drug coverage through Medicare and other types of insurance may significantly improve medication adherence and health outcomes for large numbers of chronically ill adults.

TABLE 1— Sociodemographic Characteristics and Medication Cost Burden
TABLE 1— Sociodemographic Characteristics and Medication Cost Burden
 N% of Survey Sample% of Weighted Sample
Age, y
     ≥ 6519334851
Education level
     High school or less13703455
     Some college26856645
Annual income, $
     < 20 0006951723
     20 000–39 99911612932
     40 000–59 9999802422
     ≥ 60 00012193023
Number of medications
     ≥ 79472324
Monthly out-of-pocket medication cost, $
≥ 1009742425
Prescription drug coverage
TABLE 2— Unadjusted Rates of Cost-Related Medication Underuse, Measured With Different Approaches
TABLE 2— Unadjusted Rates of Cost-Related Medication Underuse, Measured With Different Approaches
  Treatment-Specific (%)Any Medication (%)
Medical Condition% of SampleEver in the Previous YearAt Least MonthlyEver in the Previous YearAt Least Monthly
All respondents100NANA1814
High cholesterol501281814
Heart disease20952014
Previous myocardial infarction15971714
Back pain1423153430
Heart failure91081715
Chronic obstructive pulmonary disease915113028
Stomach ulcer521113629

Note. NA = not applicable. Estimates were calculated using sampling weights without adjustment for sociodemographic or clinical covariates.

TABLE 3— Logistic Regression Models Predicting Cost-Related Underuse of Specific Medication Types
TABLE 3— Logistic Regression Models Predicting Cost-Related Underuse of Specific Medication Types
 HypertensionCholesterolArthritisHeartburnDiabetes MellitusDepressionHeart DiseaseMyocardial InfarctionBack PainOsteoporosis
Age ≥ 65 ya0.2***0.3***0.5*0.3**0.2***0.4**0.1***0.1***0.70.7
Age 55–64 ya0.6****
Income < $20 000b5.9***2.7***3.9***3.4***1.63.0**6.4**5.5**5.0***4.8**
Income $20 000–$39 999b3.0***1.63.2**2.2*2.12.6*
Income $40 000–$59 999b1.
3–6 prescriptionsc0.
≥ 7 prescriptionsc1.*
Monthly prescription cost ≥ $100d4.6***3.8***2.9***5.3***1.84.5***3.8**4.4**3.7**4.5**
Monthly prescription cost $51–$99d2.0*2.3**
No prescription coveragee2.3***1.8*1.32.1**1.3
     (alternative model)f3.3***2.3***1.9*3.3***2.3*2.8***2.8*3.8**3.0**2.1

Note. Cell entries are odds ratios from multivariate models calculated using sampling weights and adjusting for all covariates shown, as well as respondents’ race (White vs other), gender, and education level.

a vs age 50–54.

b vs income ≥ $60 000/year.

c vs 1–2 prescriptions.

d vs monthly out-of-pocket cost of $0–$50.

e vs any coverage.

f Adjusted for all covariates in the primary models with the exception of out-of-pocket medication costs.

* P <. 05; **P <. 01; ***P < .001; 2-tailed tests.

This study was funded by the Health Services Research and Development Service, Department of Veterans Affairs (grant DII 99187–1). Elizabeth Eby provided data analytic support, and Joanne A. Kimata provided editorial assistance.

Human Participant Protection This study was approved by human subjects committees at the VA Ann Arbor Healthcare System and Stanford University.


1. Safran DG, Neuman P, Schoen C, Montgomery JE, Li W, Wilson IB, Kitchman MS, Bowen AE, Rogers WH. Prescription drug coverage and seniors: how well are states closing the gap? Health Aff. 2002;W253–W268. Google Scholar
2. Steinman MA, Sands LP, Covinsky KE. Self-restriction of medications due to cost in seniors without prescription coverage. J Gen Intern Med. 2001;16:793–799. Crossref, MedlineGoogle Scholar
3. Soumerai SB, Avorn J, Ross-Degnan D, Gortmaker S. Payment restrictions for prescription drugs under Medicaid: effects on therapy, cost, and equity. N Engl J Med. 1987;317:550–556. Crossref, MedlineGoogle Scholar
4. Tamblyn R, Laprise R, Hanley JA, Abrahamowicz M, Scott S, Mayo N, Hurley J, Grad R, Latimer E, Perreault R, McLeod P, Huang A, Larochelle P, Mallet L. Adverse events associated with prescription drug cost-sharing among poor and elderly persons. JAMA. 2001;285:421–429. Crossref, MedlineGoogle Scholar
5. Soumerai SB, Ross-Degnan D, Avorn J, McLaughlin TJ, Choodnovsky I. Effects of Medicaid drug-payment limits on admission to hospitals and nursing homes. N Engl J Med. 1991;325:1072–1077. Crossref, MedlineGoogle Scholar
6. Soumerai SB, McLaughlin TJ, Ross-Degnan D, Casteris CS, Bollini P. Effects of limiting Medicaid drug-reimbursement benefits on the use of psychotropic agents and acute mental health services by patients with schizophrenia. N Engl J Med. 1994;331:650–655. Crossref, MedlineGoogle Scholar
7. National Health and Nutrition Examination Survey: Patterns of Prescription Drug Use in the United States, 1988–94. Available at: http://www.cdc.gov/nchs/data/nhanes/databriefs/preuse.pdf [PDF file]. Accessed October 28, 2003. Google Scholar
8. Brown C, Schulberg HC, Sacco D, Perel JM, Houck PR. Effectiveness of treatments for major depression in primary medical care practice: a post hoc analysis of outcomes for African American and White patients. J Affect Disord. 1999;53:185–192. Crossref, MedlineGoogle Scholar
9. Krosnick JA and Chang L. A comparison of random-digit dialing telephone survey methodology with Internet survey methodology as implemented by Knowledge Networks and Harris Interactive. April 2001. Ohio State University. Available at: http://www.knowledgenetworks.com/ganp/reviewer-info.html. Google Scholar
10. Baker LC, Bundorf MK, Singer S, and Wagner TH. Validity of the survey of health and the Internet, and Knowledge Network’s panel and sampling. 2003. Palo Alto, Ca, Stanford University. Available at: http://www.knowledgenetworks.com/ganp/reviewer-info.html. Google Scholar
11. Harris KM. Can high quality overcome consumer resistance to restricted provider access? Evidence from a health plan choice experiment. Health Serv Res. 2002;37:551–571. Crossref, MedlineGoogle Scholar
12. Skitka LJ, Mullen E, Griffin T, Hutchinson S, Chamberlin B. Dispositions, scripts, or motivated correction? Understanding ideological differences in explanations for social problems. J Pers Soc Psychol. 2002;83:470–87. Crossref, MedlineGoogle Scholar
13. Bureau of Labor Statistics and the Bureau of the Census. CPS (Current Population Survey) Basic Monthly Survey. April 3, 2003. Google Scholar
14. Stata, Version 8.1. College Station, Tex: Stata Corp; 2003. Google Scholar
15. CDC National Center for Chronic Disease Prevention and Health Promotion. National Diabetes Fact Sheet. Available at: www.cdc.gov/diabetes/pubs/estimates.htm. Accessed October 28, 2003. Google Scholar
16. Morbidity and Mortality Weekly Report. Self-reported asthma prevalence among adults—United States, 2000. August 17, 2001. Available at: www.cdc.gov/mmwr/preview/mmwrhtml/mm5032a3.htm. Accessed October 28, 2003. Google Scholar
17. Piette JD, Heisler M, Wagner TH. Cost-related medication underuse: do patients with chronic illnesses tell their doctors? Arch Intern Med. In press. Google Scholar
18. Alexandander GC, Casalino LP, Meltzer DO. Patient-physician communication about out-of-pocket costs. JAMA. 2003;290:953–958. Crossref, MedlineGoogle Scholar
19. Locke SE, Kowaloff HB, Hoff RG, Safran C, Popovsky MA, Cotton DJ, Finkelstein DM, Page PL, Slack WV. Computer-based interview for screening blood donors for risk of HIV transmission. JAMA. 1992;268:1505–1506. Crossref, MedlineGoogle Scholar
20. Perrine MW, Mundt JC, Searles JS, Lester LS. Validation of daily self-reported alcohol consumption using interactive voice response (IVR) technology. J Stud Alcohol. 1995;56:487–490. Crossref, MedlineGoogle Scholar
21. Koback KA, Reynolds WM, Greist JH. Computer-administered clinical rating scales: a review. Psychopharmacology. 1996;127:291–301. Crossref, MedlineGoogle Scholar
22. Garnick DW, Hendricks AM, Thorpe KE, New-house JP, Donelan K, Blendon RJ. How well do Americans understand their health coverage? Health Aff. 1993;12:204–212. Crossref, MedlineGoogle Scholar
23. Marquis MS. Consumers’ knowledge about their health insurance coverage. Health Care Financing Review. 1983;5:65–80. MedlineGoogle Scholar


No related items




John D. Piette, PhD, Michele Heisler, MD, and Todd H. Wagner, PhDJohn D. Piette and Michele Heisler are with the Center for Practice Management and Outcomes Research, Department of Veterans Affairs and the Department of Internal Medicine and Diabetes Research and Training Center, University of Michigan, Ann Arbor. Todd H. Wagner is with the Health Economics Resource Center, Department of Veterans Affairs and the Department of Health Research and Policy, Stanford University. “Cost-Related Medication Underuse Among Chronically III Adults: the Treatments People Forgo, How Often, and Who Is at Risk”, American Journal of Public Health 94, no. 10 (October 1, 2004): pp. 1782-1787.


PMID: 15451750